The Tens-of-Billions CapexNarrow moat

The most forbidding barrier in memory, and the deepest reason the oligopoly is stable, is capital. A single leading-edge memory fabrication plant costs well over ten billion dollars¹ to build and equip, and Micron's total annual capital spending runs to tens of billions of dollars — swelling further as it races to add HBM and advanced-DRAM capacity for the AI boom. This is a sum only a handful of companies on earth can sustain, and the requirement to spend it, year after year, with no guarantee of return, is precisely what keeps the field of serious memory makers to three.

The capital barrier protects the incumbents by making entry almost unthinkable. A would-be challenger must commit tens of billions of dollars over many years, climb the brutal learning curve to competitive yields, and survive the losses of the ramp — all while three scaled incumbents, who have already paid these costs decades over, stand ready to flood the market and price the newcomer into oblivion. No amount of cleverness substitutes for the capital, and no private investor will fund losses on that scale against those odds. The barrier is why memory, for all its strategic importance, has seen no successful new Western or Korean entrant in decades. The mirror image is that this same capital intensity is a crushing burden on Micron itself — an unrelenting claim on cash borne through every cycle — and that a state willing to fund losses indefinitely can, unlike a private entrant, eventually buy its way over the wall. But as a barrier to ordinary competition, the tens-of-billions cost of admission is Micron's single strongest structural protection.